Mimo Detection And The Uncertainty In The Harq Combination Treatment

MIMO has become a milestone over the development history of mobile wireless communication technologies, with regard to its remarkable improve-ments in spectrum utilization. In practice, MIMO is often used in combination with HARQ and constitutes an MIMO-HARQ system. The existing MIMO de-tection algorithms and the combining algorithms in MIMO-HARQ systems are based on the traditional Bayesian theory. The Bayesian theory and the meth-ods based on which are important tools in measuring uncertainty information, because its theory is complete, and it can provide practical and effective formu-las for combining uncertainty measures. However, it depicts the uncertainty of knowledge deficiently, and it usually requires a priori probability or the condi-tional probability for the event, which thereby limits its wide applications. DST is an extension of Bayesian theory, which has unique merits in uncertainty pro-cessing, and the methods based on Bayesian theory are special cases of those based on DST. In the framework of DST, this paper proposes several effective MIMO detection algorithms and combining schemes in MIMO-HARQ sys-tems, which are concluded as follow:Firstly, a DS detection algorithm is proposed in MIMO systems. FES (Fo-cal Element Set) characterizes the uncertainty contained in the decision statis-tics, and the corresponding BPA of FES is the likelihood measure, acting as the soft-decisions of the transmit signals. The uncertainty can be counter-acted during combining the soft information sources from all receive anten-nas, so more reliable decisions are achieved. Moreover, a DST-based iterative detection-decoding algorithm is proposed in MIMO systems. On the one hand, the detection algorithm takes advantage of the characteristics of DST process-ing uncertainty, improving detection performance. On the other hand, during iterations the detection scheme allows for the soft information outputs from the decoder when combining multiple soft information sources, leading to the performance improvement.Secondly, MRC is regarded as the optimal combining scheme under the framework of Bayesian theory. This paper proposes a DST based combin-ing scheme for SISO-HARQ systems that outperforms MRC, referred to as DSC, in which two methods for soft information calculations are developed for equiprobable (EP) and non-equiprobable (NEP) sources, respectively. One is based on the distance from the received signal to the decision candidate set con-sisting of adjacent constellation points when the source bits are equiprobable, and the corresponding DSC is regarded as SISO-DSC-D. The other is based on the posterior probability of the transmitted signals when the priori probability for the NEP source bits is available, and the corresponding DSC is regarded as SISO-DSC-APP. For EP source, SISO-DSC-D and SISO-DSC-APP are equiv-alent in performance, and both outperform MRC. When the source is NEP and the priori probability is available to the receiver, SISO-DSC-APP makes use of the priori probability, so as to outperform SISO-DSC-D but equals MRC-MAP in performance in low SNR (Signal Noise Ratio) region, which denotes the MRC based on the MAP decision rule. When the SNR increases, the perfor-mance gap between SISO-DSC-APP and SISO-DSC-D becomes smaller, but they still outperforms MRC-MAP. Moreover, compared to MRC, the proposed DSC with merits in uncertainty processing has robustness to combat the fading channels.Thirdly, this paper proposes two DS detection-aided DS combining schemes for symbol-level and bit-level combining, referred to as DS-Symbl-DSC and DS-Bit-DSC, respectively, which are soft-detection-soft-combining schemes. In DS detection stage, uncertain decision propositions characterize the uncertainty contained in the decision statistics, and the defined symbol-level or bit-level soft information is calculated according to the received sig-nals, then act as an evidence source. In DS combining stage, combination operations to the multiple soft information sources can counteracted the uncer- tainty. Compared to the traditional MMSE detection based symbol-level post combining MMSE-post and bit-level LLR combining MMSE-LLR, both DS-Symbl-DSC and DS-Bit-DSC perform better. For the NEP source, the proposed DS-Symbl-DSC outperforms an ML combining scheme for MIMO-HARQ systems. Moreover, the bit-level DS combining is deduced to be a universal scheme, of which the LLR combining is a special case when the likelihood probability is used as bit-level soft information. Computational complexity analysis demonstrates that the proposed schemes have lower complexity.Fourthly, SA (Simulated Annealing) based MUD (Multi-User Detection) scheme is proposed in synchronous SDMA (Space Division Multiple Ac-cess) systems, which performs better and approaches the performance of ML MUD and imposes lower complexity. SA MUD is quite applicable to high-dimensional deficient MIMO detection because of its efficiency in calcula-tions. For low-dimensional deficient MIMO systems, this paper proposes an ML-MMSE detection scheme that takes advantages of ML and MMSE. The ML-MMSE scheme first picks up several sub-channels with worse SNR, on which the ML detection is applied to the transmitted signals, and the MMSE detection is applied to the signals transmitted on the other sub-channels that has better SNR. Therefore, the proposed ML-MMSE can achieve a good tradeoff between performance and calculation complexity.